Some research projects may have been funded more than once, according to a new analysis of over 600,000 grant and contract summaries from the largest US funders. Overall, tens of millions of dollars were awarded to projects that appear to have been already either partially or completely funded.
The analysis was conducted on a supercomputer built by Garner, called Shadowfax, which is typically used for data mining, genomics, and similar applications. The grant applications were drawn from the five largest funders in the United States: Susan J. Komen for the Cure, Department of Defense, Department of Energy, National Science Foundation, and the National Institutes of Health.
Over a period of three months, Shadowfax conducted about 200 billion textual comparisons between the grant summaries to identify similar or duplicate projects. Then the team filtered the data to remove applications that looked like they were connected to conferences or large projects that would require funding from multiple agencies.
While the computer immediately identified several thousand possible duplications, the analysis required further human review because unique grant applications from the same research group often contains similar wording and research background sections. So Garner personally reviewed every grant pair the computer flagged as a potential duplicate.
In the end, Garner and colleagues identified 167 pairs of grant summaries that they believed were likely to have been double funded. Between 2007 and 2011, the authors identified 39 pairs that totaled more than $20 million in research funding.
In a time of fierce competition for research funding, agencies could do a similar analysis to flag potential double-dipping grant applications for further review. “It could help detect this before [funding] happens, or to make adjustments to various budgets to compensate for overlap,” said Garner.
Garner is convinced that the findings greatly underestimate the amount of double-dipping that exists. “We know from surveys of scientists (2), that a large fraction, maybe a couple of percent, admits to having plagiarized [in papers]. We found only a tenth of a percent or so, so we know there is about 35 times more plagiarism being done than we can discover with our software,” he said.
If a similar number of researchers are dishonest in their grant applications, the amount of duplicate funding could be as much as $5.1 billion since 1985, or about 2.5% of all funds, Garner estimates. In biomedicine alone, Garner estimates the amount of money spent on double-funded projects since 1985 to be about $300 million. “There is a lot of research that could be funded from money saved if double dipping is curtailed,” he said.
And his motives aren’t entirely selfless: “I would love to get more grants,” he said.
1. Garner, H.R., L.J. McIver, and M.B. Waitzkin. 2013. Nature 493:599-601.
2. Martinson, B.C., M.S. Anderson, and R. de Vries. 2005. Nature 435: 737–738.